Dear FSL developers and users,
I would like to hear your opinions, suggestions or to get pointers to
relevant references concerning removal of subject movement related
artefacts using MELODIC. I have some patient data with movement
artefacts which are partly correlated with the experimental setup. The
main problem is that MELODIC gives a large number (> 200) independent
components. I can recognise that some are clearly movement related but
there are also less clear cut cases.
I wonder whether there would be suggestions for some heuristic or metric
to identify the movement related components for denoising. Or should I
limit the number of components? I learned from the documentation that
this is not advised or fMRI data.
Thank you in advance,
Antti Korvenoja
|